TELEGRAPH (from rilte and ypicfra)) signifies an instrument to write at a distance. The term is specifically applied to apparatus for communicating intelligence to a .distance in unwritten signs addressed to the eye or ear, and has only recently had application to those wonderful combinations of inanimate matter which literally write at a distance the intelligence committed to them. The chief object of the present article is to explain the principles and practice of the electric telegraph, and we shall allude to other telegraphic systems only to illustrate the general principles of signalling.

A word expressing an idea may, according to a prearranged plan of signalling, be communicated by voice, by trumpet calls, by gun fire, by gesture or dumb signs, by lamp signals, by flags, by semaphore, or by electric telegraph. The simplest system of word-signalling hitherto practised is that of the nautical flag telegraph, in which each hoist represents a word by a combination of four flags in four distinct positions (see SIGNALS, NAVAL). If n denote the number of flags, supposed all different, out of which the four to be sent up may be selected, the number of different ideas which can be expressed by a single hoist is n(n - 1)(n - 2)(n - 3), since there are n varieties out of which the flag for each of the four positions may be independently chosen. To commit to memory so great a number of combinations, which amount to 358,800 if n = 26, would be a vain effort ; the operators on each side must therefore have constant recourse to a dictionary, or code, as it is called. For the sake of convenient reference each flag is called by the name of a letter of the alphabet, and all that the operator has to bear in mind is the letter by which each flag is designated. Sometimes the words to be expressed are spelled out by means of the flags as in ordinary language ; but, as in most words there are more than four letters, as scarcely any two consecutive words are spelled with four or less than four letters, and as more than four flags at a time cannot be conveniently used, the system of alphabetic signalling frequently requires the use of two hoists for a word, and scarcely ever has the advantage of expressing two words by one hoist. It is therefore much more tedious than code signalling in the nautical telegraph.

In point of simplicity spoken words may be considered as almost on a par with the nautical telegraph, since each word is in reality spoken and heard almost as a single utterance. Next in order comes the systein of spelling out words letter by letter, in which - instead of, as in the nautical telegraph, 358,800 single symbols to express the same number of ideas-26 distinct symbols are used to express by their combinations any number whatever of I See his Mathematical and Physical Papers, vol. ii. p. 105.

distinct ideas. Next again to this may be ranked the system by which several distinct successive signals are used to express a letter, and letters thus communicated by compound signals are combined into words according to the ordinary method of language. It is to this last class that nearly all practical systems of electro-telegraphic signalling belong. But some of the earliest and latest proposals for electric telegraphs are founded on the idea of making a single signal represent a single, letter of the alphabet ; as instances we may name those early forms in which separate conductors were used for the different letters ; a method suggested by Professor W. Thomson ' in 1858 in which different strengths of current were to be employed to indicate the letters ; and the various forms of printing telegraph now in use.

I. HISTORICAL SKETCH OF EARLY TELEGRAPHS.

Although the history of practical electric telegraphy Early does not include a period of more than half a century, the fmins• idea of using electricity for telegraphic purposes is much older. It was suggested again and again as each new discovery in electricity and magnetism seemed to render it more feasible. Thus the discovery of Stephen Gray and of Wheeler that the electrical influence of a charged Leyden jar may be conveyed to a distance by means of an insulated wire gave rise to various proposals, of which perhaps the earliest was that in an anonymous letter to the Scots Magazine (vol. xv. p. 73, 1753), in which the use of as many insulated conductors as there are letters in the alphabet was suggested. Each wire was to be used for the transmission of one letter only, and the message was to be sent by charging the proper wires in succession and received by observing the movements of small pieces of paper marked with the letters of the alphabet and placed under the ends of the wires. A very interesting modification was also proposed in the same letter, viz., to attach to the end of each wire a small light ball which when charged would be attracted towards an adjacent bell and strike it. Some twenty years later Le Sage proposed a similar method, in which each conductor was to be attached to a pith ball electroscope. An important advance on this was proposed in 1797 by Lomond,3 who used only one line of wire and an alphabet of motions. Besides these we have in the same period the spark telegraph of Reiser, of Don Silva, and of Cavallo, the pith ball telegraph of Ronalds, and several others. Next came the discovery of Galvani and of Volta, "ABC instruments," were worked out with great ingenuity and as a consequence a fresh set of proposals, in which of detail by Wheatstone in Great Britain and by Breguet voltaic electricity was to be used. The discovery by and others in France. They are still largely used for priNicholson and Carlisle of the decomposition of water and vete wires, but are being rapidly displaced by the telephone.2 the subsequent researches of Davy on the decomposition of Wheatstone also described and to some extent worked out the solutions of salts by the voltaic current were turned to an interesting modification of his step-by-step instrument, account in the water voltameter telegraph of Sommering the object of which was to produce a letter-printing teleand the modification of it proposed by Schweigger, and in graph. But it never came into use ; some years later, a similar method proposed by Coxe, in which a solution of however, an instrument embodying the same principle, salts was substituted for water. Then came the discovery although differing greatly in mechanical detail, was brought by Romagnesi and by Oersted of the action of the galvanic into use by Royal E. House of Vermont, U.S., and was current on a magnet. The application of this to tele- very successfully worked on some of the American telegraph graphic purposes was suggested by Laplace and taken up lines till 1860, after which it was gradually displaced by by Ampere, and afterwards by Triboaillet and by Schilling, the Phelps combination telegraph. The House instrument whose work forms the foundation of much of modern tele- is not now in use, but various modifications of it are still graphy. Faraday's discovery of the induced current pro- employed for private lines and for stock telegraphs, such duced by passing a magnet through a helix of wire forming as Calahan's and the universal stock telegraphs, Phelps's part of a closed circuit was laid hold of in the telegraph of stock printer, Gray's automatic printer for private lines, Gauss and Weber, and this application was at the request Siemens's and Phelps's automatic type printers, &c. (see of Gauss taken up by Steinheil, who brought it to consider- infra, pp. 120-121).

able perfection. Steinheil communicated to the Glittingen II. GENERAL DESCRIPTION OF ELECTRIC TELEGRAPHS Academy of Sciences in September 1838 an account of his telegraph, which had been constructed about the middle of the preceding year. The currents were produced by a The first requisite for electro-telegraphic communica- Essential magneto-electric machine resembling that of Clarke. The tion between two localities is an insulated conductor ex- apparareceiving apparatus consisted of a multiplier, in the centre tending from one to the other. This, with proper apparatuS tn.% of which were pivoted one or two magnetic needles, which for originating electric currents at one end and for diseither indicated the message by the movement of an index covering the effects produced by them at the other end, or by striking two bells or different tone or recorded it by constitutes an electric telegraph. Faraday's term " eke-making ink dots on a ribbon of paper. Among other trode," literally a way for electricity to travel along, might workers about this time we may mention Masson, Breguet, be well applied to designate the insulated conductor along Davy, Deval, Billon, Soudalot, and Vorsselman who pro- which the electric messenger is despatched. It is, how-posed to use the physiological effects of electricity in work- ever, more commonly and familiarly called "the wire" or ing an electric telegraph.1 "the line."

Steinheil appears to have been anticipated in the matter The apparatus for generating the electric action at one of a recording telegraph by Morse of America, who in 1835 end is commonly called the transmitting apparatus or in-constructed a rude working model of an instrument; this strument, or the sending apparatus or instrument, or some-within a few years was so perfected that with some modi- times simply the transmitter or sender. The apparatus fication in detail it has been largely used ever since (see used at the other end• of the line to render the effects of below). In 1836 Cooke, to whom the idea appears to have this action perceptible to any of the senses - eye, ear, or been suggested by Schilling's method, invented a telegraph taste, all of which have been used in actual telegraphic in which an alphabet was worked out by the single and signalling - is called the receiving apparatus or instrument.

may consult Edward Highton, The Electric Telegraph, London, 1852; Moigno, Traite de Tglegraphie Electrique, Paris, 1849; and Sabine, 2 For the different forms, see Prescott's Electricity and the Electric .history of the Electric Telegraph, London, 1869. Telegraph, pp. 562-602.

- is shown in fig. 1. It consists of two distinct cups (e, C), which are moulded and fired separately, and afterwards cemented together. The double cup gives great security against loss of insulation due to cracks extending through the insulator, and also gives a high surface insulation. An iron bolt (b) cemented into the centre of the inner cup is used for fixing the insulator to the pole or bracket.

Under- In the underground system the main ground line generally consists of a copper wire, lines' or a thin strand of copper wires, covered with a continuous coating of gutta percha, india-rubber, or some equivalent insulating substance, served with tarred tape and enclosed in earthenware, iron, or lead Id pipes laid below the surface of the ground.

This system is largely used for street and tunnel work, and to a considerable extent, especially in Germany, for ordinary lines. Fia. 1. - Varley's Each tube generally contains a number of ctioornholene_fopuroisgii wires, which are either laid up into a cable size.

and covered with a serving of tarred tape or hemp before being drawn into the tube, or - as is more commonly the case in the United Kingdom - simply laid together in a parallel group and tied at intervals with binders, which are removed as the wires are drawn into the tube. On some long underground lines in Germany the insulated wires are laid up into a cable, served with jute or hemp, and sheathed with a continuous covering of iron wires, precisely similar to the submarine cables described below. The cable is laid in a deep trench and coated with bitumen. This form of cable is easily laid, and if properly manufactured is likely to be very durable.

Sub- Submarine Cables. - A submarine cable (figs. 2-4), as marine usually mann.- present, con- 101-191Nit sists of a core a in the centre of which is a strand of copper wires varying in weight for different cables between 70 and 400 lb to the mile. The stranded form was sug- W. Thomson 116#1111°11 gested by Prof. at a meeting of the Philo- P1301111 bility renders 4I511105 it less likely kiatr- Fig. 4.

to damage the insulating en- Fig. 3.

velope during 28-14z.e - . Seleitig.ols of three tomes oil' nse. etblteers: - Type the manipula- mediate type. Fig. 4. - Deep sea type.

tion of the cable. The central conductor is covered with several continuous coatings of gutta percha, the total weight of which also varies between 70 and 400 lb to the mile. With a light core the weight of the gutta percha generally exceeds that of the copper, while in some heavy cores the copper is heavier. The different coatings of gutta percha and of the conductor are usually separated by a thin coating of Chatterton's compound (a mixture of gutta percha, resin, and Stockholm tar), in order to make them adhere firmly together. This practice has recently been departed from by Messrs Siemens Brothers, who have succeeded by an improved process of manufacture in getting perfect adhesion without the use of the compound. The core is served with a thick coating of wet jute, yarn, or hemp (It), forming a soft bed for the sheath, which consists of soft iron, or of homogeneous iron, wires of the best quality. The sheathing wires are usually covered with one or two servings of tarred canvas tape (t), or of tarred hemp, laid on alternately with coatings of a mixture of asphaltum and tar. The weight of the iron sheath varies greatly according to the depth of the water, the nature of the sea bottom, the prevalence of currents, and so on. Fig. 2 shows the intermediate type again sheathed with a heavy armour to resist wear in the shallow water near shore. In many cases a still heavier type is used for the first mile or two from shore, and several intermediate types are often introduced, tapering gradually to the thin deep-water type. Captain S. Trot and Mr F. A. Hamilton have proposed 1 to abandon the iron sheath and substitute a strong double serving of hemp, laid on in such a way as to prevent twisting when the cable is under tension. This suggestion, which is a revival with some modifications of an old idea, is, however, still in the experimental stage.

We will now describe very briefly a few of the most important processes in the manufacture arid submergence of submarine cables.

In manufacturing a cable (fig. 5) the copper strand is passed Their through a vessel A containing melted Chatterton's compound, then manuthrough the cylinder C, in which a quantity of gutta percha, purl- facture.

fled byre eated wash- ing in hot water, by mastication, and by filtering through wire-gauze filters, is kept warm by a steam- jacket. As the wire is pulled through, a coating of gutta percha, the thickness of A which is regulated by the die D, is pressed out of the cylinder by applying the requisite pressure to the piston P. The newly coated wire is passed through a long trough T, containing cold water, until it is sufficiently cold to allow it to be safely wound on a bobbin B'. This operation completed, the wire is wound from the bobbin B' on to another, and at the same time carefully examined for air-holes or other flaws, all of which are eliminated. The coated wire is treated in the same way as the copper strand, - the die D, or another of the same size, being placed at the back of the cylinder and a larger one substituted at the front. A second coating is then laid on, and after it passes through a similar process of examination a third coating is applied, and so on until the requisite number is completed. The finished core changes rapidly in its electric qualities at first, and is generally kept for a stated interval of time before being subjected to the specified tests. It is then placed in a tank of water and kept at a certain fixed temperature, usually 75° Fahr., until it assumes approximately a constant electrical state. Its conductor and dielectric resistance and its electrostatic capacity are then measured. These tests are generally repeated at another temperature, say 50° Fahr., for the purpose of obtaining at the same time greater certainty of the soundness of the core and the rate of variation of the conductor and dielectric resistances with temperature. Should these tests prove satisfactory the core is served with jute yarn, coiled in watertight tanks, and surrounded with salt water. The insulation is again tested, and if no fault is discovered the served core is passed through the sheathing machine, and the iron sheath and the outer covering are laid on. As the cable is sheathed it is stored in largo water-tight tanks and kept at a nearly uniform temperature by means of water.

The cable is now transferred to a cable ship, provided with water- Submertight tanks similar to those used in the factory for storing it. The sion. tanks are nearly cylindrical in form and have a truncated cone must therefore be so regulated coils are prevented from adhering by a coat- that the angle of immersion is as great as the inclination of the c T ing of whitewash, and the end of each steepest slope passed over. Under ordinary circumstances the angle nautical mile is carefully marked for future of immersion i varies between six and nine degrees.2 found to be in perfect condition, the ship is taken to the place where the shore end is to be landed. A sufficient length of cable to reach the shore or the cable-house is paid overboard and coiled room in the cable-house and the conductor connected with the Qualities of a Telegraph Ling. - The efficiency of the telegraph testing instruments, and, should the electrical tests continue satis- depends on three qualities of the main line - (1) its conducting factory, the ship is put on the proper course and steams slowly ahead, quality, (2) its insulation, and (3) its electrostatic capacity.

which it leaves the ship and for measuring the pull on the cable. be most naturally, and is in point of fact generally, expressed in The essential parts of this apparatus are shown in fig. 6. The lower terms of resistance to transmission, regarded as a quality inverse to end a of the cable in the tank T is taken to the testing room, so that that of conducting power, and expressed numerically by the reel-continuous tests for electrical condition can he made. The upper procal of the measure of the conducting power. An independent end is passed over a guiding quadrant Q to a set of wheels or fixed explanation and definition of the electrical resistance of a conductor quadrants 1, 2, 3, ... then to the paying- out drum P, from it to may be given as follows : - the electrical resistance of a conductor the dynamometer D, and finally to the stern pulley, over which it is measured by the amount of electromotive force, or difference of passes into the sea. The wheels 1, 2, 3, ... are so arranged that potentials, which must be maintained between its ends to produce 2, 4, 6, ... can be raised or lowered so as to give the cable less or a stated strength of electric current through it.

which it passes several times to prevent slipping. On the same insulation. Since no substance yet known is absolutely a non-shaft with P is fixed a brake-wheel furnished with a powerful conductor of electricity, perfect insulation is impossible. If, brake B, by the proper manipulation of which the speed of paying however, the supports on which a telegraph wire rests present, on out is regulated, the pull on the cable being at the same time each part and on the whole; so great a resistance to electric conducobserved by means of D. The shaft of P can be readily put in gear tion as to allow only a small portion of the electricity sent in, in the with a powerful engine for the purpose of hauling back the cable actual working, at one end to escape by lateral conduction, instead should it be found necessary to do so. The length paid out and of passing through the line and producing effect at the other end, the rate of paying out are obtained approximately from the number the insulation is as good as need be for the mode of working adopted. of turns made by the drum P and its rate of turning. This is With the good insulation attained in a submarine line, round every checked by the mile marks, the known position of the joints, &c., part of which the gutta percha is free from flaws, no telegraphic as they pass. The speed of the ship can he roughly estimated operation completed within a second of time can be sensibly from the speed of the engines ; it is more accurately obtained influenced by lateral conduction. A charge communicated to a by one or other of the various forms of log, or it may be measured wire thus insulated under water, at the temperature of the sea-by paying out continuously a steel wire over a measuring wheel. bottom, is so well held that, after thirty minutes, not so much as The average speed is obtained very accurately from solar and stellar . half of it is found to have escaped. From this, according to the observations for the position of the ship. The difference between familiar "compound interest" problem,• it appears that the loss the speed of the ship and the rate of paying out gives the amount must be at the rate of less than five per cent. per two minutes.

slack is not a guarantee that the cable will always lie closely along potentials between the conductor at any point and the earth beside the bottom or be free from spans. Whilst it is being paid out the it. In 1854 Faraday showed the effect of this "electrostatic charge" portion between the surface of the water and the bottom of the sea on signals sent through great lengths of submerged wire, bringing lies along a straight line, the component of the weight at right to light many remarkable phenomena and pointing out the angles to its length being supported by the frictional resistance to " inductive " embarrassment to be expected in working long sub-sinking in the water. If, then, the speed of the ship be v, the marine telegraphs. In letters3 to Professor Stokes in November rate of paying out u, the angle of immersion. i, the depth of the and December of the same year, Prof. W. Thomson gave the mathewater h, the weight per unit length of the cable so, the pull on the matical theory of these phenomena, with formulae and diagrams of cable at the surface P, and A, B constants, we have - curves, containing the elements of synthetical investigation for A every possible case of practical operations. Some of the results of P=Tilw- (u - v cos i') SID z ((fla)), this theory are given at the end of the present article. The conductor of a submarine cable has a very large electrostatic capacity and w cos i=Bf (v sin i) in comparison with that of a land telegraph wire in consequence of where f stands for "function." The factors Af (u - v cos i) and the induction, as of a Leyden phial, which takes place across its Bf (v sin i) give the frictional resistance to sinking, per unit length gutta percha coat, between it and its moist outer surface, which of the cable, in the direction of the length and transverse to the may he regarded as perfectly connected with the earth, - that is to length respectively.' It is evident from equation (9) that the say, at the same potential as the earth. The mathematical expresangle of immersion depends solely on the speed of the ship ; sions for the absolute electrostatic capacity C, per unit of length, hence in laying a cable on an irregular bottom it is of Feat ins- in the two cases are as follows.

portance that the speed should be sufficiently low. This may be Let D = diameter of the inner conductor, supposed to he that of a Submarillustrated very simply as follows : - suppose a a (fig. 7) to be the circular cross section, or of a circle inappreciably less than one cir- inc line. surface of the sea, bc the bottom, and cc the straight line made cumscribed about the strand which constitutes a modern submarine by the cable ; then, if a hill H, which is at any part steeper than Electric Telegraph Cables, London, 1678.

i See Sir W. Thomson, Mathematical and Physical Papers, voL IL p. 165. 3 Published in Proc. Roy. Soc. for 1855.

conductor ; D'= outer diameter of the insulating coat; I = specific inductive capacity of the gutta percha or other substance constituting the insulating coat. Then Air line. In the case of a single wire of circular section, diameter D, un-2 log, 4h/D' Example 1. In a submarine cable in which D'= 1 centimetre ; D =0.4 centimetre; and I=3.2 - 3.2 x .4343=115.

C =21og, D'/D - 2 x .3979 C _2 log, 4h/D 7.204 16.6' The capacity, therefore, is in this case less than one-twenty-ninth of that of the submarine cable of example 1 for the same length.

Telegraph Testing.

Stand- Standards of Measurement. - A brief consideration of the standards arch of according to which the electrical qualities referred to in the last measure- section are measured, and the measurements to be described in this ment. section are made, will render the statements of those qualities and quantities more definite. A complete and universally comparable system of standards for physical measurements can be obtained by adopting arbitrarily as fundamental units those of length, mass, and time, and expressing in terms of these in a properly defined manner the units of all the other quantities. The units now adopted all over the world for electrical measurements take the centimetre as the unit of length, the gramme as the unit of mass, and the mean solar second as the unit of time. There are two systems in use, the electrostatic and the electromagnetic. In the former the mutual forces exerted by two bodies, each charged with static electricity, are taken as the starting-point, and in the latter the mutual forces exerted between a current of electricity and a magnet. The units according to these two systems are definitely related ; but as we deal in the present article with the electromagnetic system we give the following brief account of it only.

Units. The dyne or unit force is that force which, acting on a gramme of matter, free to move, imparts to it a velocity of 1 centimetre per second. Unit quantity of magnetism or unit magnetic pole is that quantity of magnetism which, when placed at a distance of 1 centimetre from an equal and similar quantity of magnetism or a magnetic pole, repels it with unit force. Unit magnetic field is a field which, when a unit quantity of magnetism or a unit magnetic pole is placed in it, is acted on by unit force. Unit current is a current which, when made to flow round a circle of unit radius, produces a magnetic field of 2,r units' intensity at the centre of the circle, or acts on a unit quantity of magnetism placed at the centre of the circle with 2w units of force. Unit quantity of electricity is the quantity conveyed by the unit current in one second. Unit difference of potential is the difference of potential between the ends of a conductor of unit length when it is placed with its length at right angles to the direction of force in a unit magnetic field and kept moving with a velocity of 1 centimetre per second in the direction at right angles to its own length and to the direction of the magnetic force. Unit electromotive force is produced in a closed circuit if' any unit of its length is held in the manner, and moved in the direction and with the velocity, described in the last section. Unit resistance is the resistance which, when acted on by unit electromotive force, transmits unit current. Unit capacity is the capacity of a body which requires unit quantity of electricity to raise its potential by unity. The units above specified are generally referred to as the absolute C.G.S. electromagnetic units of the different quantities. In practice their magnitudes were found inconvenient, and certain multiples and submultiples of them have been adopted as the practical units of measurement : thus the ohm is equal to 10' C.G.S. units of resistance ; the volt is equal to 108 C.G.S. units of electromotive force ; the ampere is equal to 10-1 C.G.S. units of current • the coulomb is equal to 10-1 C.G.S. units of quantity ; the farad is the capacity which is charged to a volt by a coulomb, and is equal to 10-8 C.G.S. units of capacity ; the inicrofarad is the millionth part of the farad, and is equal to 10-1' C.G.S. units of capacity. - We are here chiefly concerned with the units of electromotive force, resistance, and capacity. No universally recognized standard of electromotive force has yet been established, but the want has been to a great extent supplied by the potential galvanometers, electrostatic voltmeters, standard cells, and other instruments devised by. Sir W. Thomson and others. The work of Lord Rayleigh, Dr. Fleming, and other experimenters on the Clark and Daniell standard cells has shown conclusively that an electromotive force can be reproduced with certainty within one-tenth per cent. of accuracy by means of either of these cells. Specimens of the standard unit of resistance, or ohm, made of an alloy of platinum and silver, or of platinum and iridium, have been constructed, and can be relied on, if properly taken care of, to remain very nearly accurate from year to year. Similar specimens of the standard unit of capacity or microfarad which remain very nearly constant have been successfully produced. For a fuller treatment of this subject and of the methods of determining the different units, see ELECTRICITY, Vol. viii. p. 40 sq.' Telegraph line testing consists mostly of comparisons of the resistance of the conductor and the insulator with sets of standard resistances, and of comparisons of the inductive capacity of the line or cable with standard condensers of known capacity. When, as is sometimes the case, the strength of the current flowing through the line or through a particular instrument is to be determined, it is measured by an electrodynamometer, or by a current galvanometer, properly constructed for indicating currents in absolute measure. In the absence of such an instrument it may be obtained accurately by the use of a standard galvanometer in a known or determined magnetic field, or, taking advantage of Faraday's discovery of the electro-chemical equivalents, by measuring the amount of silver or of copper deposited by the current when it is made to pass through an electrolytic cell ; or the electromotive force per unit resistance of the circuit may be determined by the use of standard resistances and a standard cell. Space does not allow us to do more than simply refer to these methods, the first two at least of which involve accurate and somewhat difficult experimental work.' Measurement of Wire _Resistance. - (1) By Wh,eatstone's bridge.a measurex a resistance which can be varied, a sistance ; G a galvanometer, K a single lever by key, K1 a reversing key, and B a K eWheatbattery. Put the zinc pole of the stone's battery to the line and adjust the resistance x until the galvanometer G shows no deflexion when Ki is depressed. We then have, assuming no electromotive Fig. 8.

force in the line, / =axilb. Next put the copper pole to the line and repeat the test, and suppose in this casel=axdb ; if these two values of /nearly agree the true value may be taken as 2ax1xs/b(x1 + x2). The effect of an electromotive force in the line itself is nearly eliminated by reversing the battery.

(2) Let the battery B (fig. 9) be connected through the keys K1 By direct and K and the galvanometer G with the line 1, which has its distant deflexion. end to the earth as before; shunt the galvanometer by a shunt s until a convenient deflexion is obtained, and then take as quickly as possible a series of readings with zinc and copper alternately to the line. Next substitute for 1 a set of resistance coils and vary the resistance until the same series of readings is obtained. The resistance introduced for the reproduction of each reading indicates the apparent resistance of the line when that reading was taken. The readings will generally differ because of the existence of a variable electromotive force in the line, If, however, the difference is not very great, the harmonic mean of the arithmetic mean of the For the development of this important part of electrical science, see Weber, "Messungen galvanischer Leitungswiderstande nach einem absoluten Maasse," in Poggendorf s ,4nnalen, March 1851; Thomson, "Mechanical Theory of Electrolysis," " Application of the Principle of Mechanical Effect to the Measurement of Electromotive Force, and of Galvanic Resistances in Absolute Units," and "Transient Electric Currents," in Phil. Mag., 1851 and 1852; Weber, Electrodynamische Moassbestimmungen, insbesondere Zuriicilfthrung der Stromintensitiitsmessungen asif mechanischen Maass, Leipsie, 1856 ; Thomson, " On the Electric Conductivity of Commercial Copper," "Synthetical and Analytical Attempts" on the same subject, and "Measurement of the Electrostatic Force between the Poles of a Daniell's Battery, and Measurement of the Electrostatic Force required to produce a Spark in Air," Proc. Roy. Soc., 1857 and 1860; reprint of Reports of Brit. Assoc. Committee on Electr. Stand., &e., edited by Prof. F. Jenkin ; Thomson, Electrical Units of Measurement, a lecture delivered at Institution of Civil Engineers, 1883 ; Reports of the International Conference for the Determination of the Electrical Units, held at Paris in 1882 and 1884 ; A. Gray, Absolute Measurements in Electricity and Magnetism, Loudon, 1884.

To determine the constant _ a of the electrometer, con4 nett the earth wire w with a directions of the current through the slide ; its value multiplied by 10, when the slide is made up of ten coils, gives the value in scale divisions of the full difference of potential between the ends of the slide. This number added to the zero reading of the electrometer is called the inferred zero. To find the insulation of the cable, remove the wire w, put in the short circuit plug p, move the slider to contact 10, and, the distant end of the cable being insulated, apply by means of K, the zinc pole of the battery to the cable and the copper pole to the earth. Allow sufficient time for the cable to charge - say one minute for a cable of 2000 knots - then remove the short-circuit plug and take readings every fifteen or thirty seconds. The difference of these readings from zero gives the fall of potential of the cable due to discharge through the insulating coat. Next earth the cable at both ends for a time equal to the duration of the last test, and after reversing K put the copper pole of the battery to the cable and the zinc pole to the earth and take another series of readings. Subtract these readings from the inferred zero, and, using the differences as ordinates and the corresponding times as abscissae, draw two curves. To find. the insulation of the cable at any interval t after the battery was applied, draw a tangent to the curve at the point corresponding to that time and produce it to cut the axis of the ordinates. Let D1 be the ordinate to the point of intersection, and D the ordinate at the time t ; then, if C be the capacity of the cable in microfarads and I its insulation in megohms, I =C(D, - If the difference between the reading and the inferred zero at the times t and 4 be D and DI, the insulation is given by the equation scale to one side, and J__ hence the total deflexion, i illi and therefore the sensi- PI- -Trl 1-71 Wily, may be made very - considerable. In this caseI - the reversing key K is re- K, K quired for keeping the de- EN Fig. 9.

flexion in the same direc- EMI tion. With a perfectly insulated battery this can be accomplished by putting the galvanometer between the battery and the key K; but the arrangement shown is safer. The most suitable galvanometer for these tests is a dead-beat mirror galvanometer with a long enough suspension to prevent error from the viscosity of the fibre. Such an instrument is much to be preferred to the astatic form, especially when variable earth-currents are present.

By differ- (3) A highly sensitive modification of method (2) is obtained by eutial the use of a differential galvanometer, one coil of which is joined in galvano- circuit with the standard resistances and the other coil with the meter. line. The resistances are then adjusted to balance, or to give no permanent deflexion when the battery circuit is closed. Several balances with positive and negative currents must be taken and the results combined as indicated above.

By elec- (4) When an electrometer is employed for testing insulation, as trometer. described below, it may be used for the wire resistance also either by substituting it for the galvanometer in Wheatstone's bridge method (fig. 8, G) or by that shown in fig. 10. One pole of the battery B is joined to the line through the reversing key K and the resistance R, the other pole being to the x 1, noted. The deflexions should be as nearly as possible equal; that is, R should be as nearly as possible equal to 1. The form of reversing key shown at K1 is convenient for this test, as it allows the comparisons to be made quickly ; and, as the readings can be always taken to the same side of zero, the whole length of the scale is available for each deflexion. The key consists of two ordinary front and back stop single lever keys fixed together by an insulating piece i at such a distance apart that the contact stops a, b and c, d are at the corners of a square. Suppose one pole of the battery put to the line and the resistance R adjusted until no change of deflexion is obtained by depressing K1; then R is equal to 1 if there is no earth disturbance. Then put the other pole of the battery to the line ; turn the levers of K through 90° round the pivot p ; and repeat the adjustment of R for a second determination of 1. Repeat these measurements several times and combine the results in the manner described in method (2). If R is not made equal to 1, the resistances are in the ratio of the corresponding deflexions.

.reasure- Measurement of Insulator Resistance. - (1) In the direct deflexion ment method the connexions are the same as those shown in fig. 9, except of insula- that the distant end of the line is insulated. Very great care must for re- be taken that the galvanometer and all the connexions between it sistance ; and the end of the line are so well insulated that no sensible part direct of the observed deflexion is due to leakage through them. In C = Rd .

s Electro- (2) The electrometer method is only applicable to lines of conI= C log D/D,' when 4 - t is reckoned in seconds. This latter is the formula commonly used ; it gives the insulation at some time in the interval between the two observations ; the exact time depends on the rate of " absorption " of the cable.

The advantages of the electrometer method of testing cables are the comparative steadiness of the needle during earth-current disturbances, its high sensibility for the detection of small intermittent faults, and the fact that simultaneous tests can be taken from both ends of the cable. In order to test from both ends simultaneously one or other of the following methods may be adopted. Call the ends of the cable A and B, and suppose the operator at A is to begin the test. The operator at B joins the copper pole to the earth and the zinc pole to the line, and leaves the slider of his slide resistance at the earth end of the slide. Then, at a time previously arranged, he watches until he sees the electrometer begin to indicate a charge in the cable, and moves the slider along the slide so as to keep the electrometer near zero. As soon as the electrometer ceases to indicate increase of charge he ceases to move the slider and begins to record the deflexions at regular intervals, the first reading being taken as zero. The other method is to leave the slider permanently to earth and keep the electrometer so insensitive that the deflexion is always within the limits of the scale. Observe the time at which the electrometer begins to be deflected, and from that time onward take readings every thirty seconds during the time of the test. The mean of the readings taken at both ends, reduced to the same sensibility, should be used for calculating the insulation. This method not only eliminates the effects of earth-current disturbance but also throws light on the nature and distribution of such disturbances.

When an electrometer is not available and the line is too much Fall of disturbed for good tests to be obtained by the galvanometer method, potential the following procedure may be adopted. Jom the battery and the method galvanometer in series with the cable as for the direct deflexion by gal-test. Short-circuit the galvanometer and charge the cable for one vanominute. Insulate the cable for fifteen seconds ; then break the meter. short circuit of the galvanometer ; again apply the battery, and take the deflexion produced by the charge. Keep the battery on the cable for fifteen seconds, and during that time take if possible the direct deflexion reading two or three times. Again insulate for fifteen seconds and repeat the above readings ; and continue the whether it continues in perfect electrical condition, so that should Tests of same cycle of operations for the whole time of the test. After any fault develop it can be at once detected and further paying out sub-earthing the cable for the proper interval repeat the above test stopped until it is removed. It is also of great importance that merged with the other pole of the battery to the cable. To reduce the the ship and shore should be in telegraphic communication with cable.

charge readings to absolute measure, find the deflexion of the gal- each other. The arrangements made for these purposes by different vanometer needle due to the charge of a condenser of is microfarads electricians vary considerably ; but the general principle will be capacity by the testing battery • let d be this deflexion. Then the gathered from fig. 13, which includes all that is absolutely necessary deflexion that would be obtained by charging the whole cable would for the purpose. The be Cd/n, and, if D be any one of the deflexions during the test, principal testing sta- ..,.........--......

the fifteen seconds immediately preceding this charge ; thus the ship, and from it all s